Connect Claude, Cursor, opencode, or Codex to your ObserveOne account over MCP. Ask about monitor status, uptime, and incidents in plain English. Also ships as a CLI and an agent skill.
Read your monitors, uptime, and incidents, and safely create, update, and run them, from any MCP-compatible agent.
Manage monitors, run exports, and apply config-as-code from the terminal or a CI script.
Teaches any skills.sh-compatible agent how to set up the CLI (API key) or connect via MCP (OAuth), then read and manage your monitors.
ObserveOne runs a hosted MCP server. It uses OAuth; your client handles the sign-in flow. Once connected, your agent gets tools to read monitor status, uptime, and incidents, and to safely create, update, pause, and run them. There are no delete tools.
See the MCP setup docs for per-client instructions.
The obs CLI manages URL monitors, API checks, heartbeats, status pages, incidents, and Autopilot suites. Use it directly, wire it into a CI script, or let your agent call it via a bash tool.
# Installnpm install -g @observeone/cli# Authenticateobs login# Export your current monitors to obs.jsonobs export# Create a URL monitorobs url-monitor create \--name "Production API" \--url "https://api.example.com/health" \--interval "*/5 * * * *"
Full reference on the CLI page.
The observeone-agent skill tells any skills.sh-compatible agent how to connect to ObserveOne. For the CLI path, it installs @observeone/cli and asks you for an API key. For MCP, it points the agent at the server and OAuth handles auth. Either way, the agent can then read and manage your monitors through the CLI or query monitoring data over MCP.
# Works with any skills.sh-compatible agent, including Claude Code:npx skills add Observeone1/observeone-agent# The skill guides the agent to:# 1. Install @observeone/cli and ask you for an API key# (or connect via MCP with OAuth instead)# 2. Use the CLI to manage your monitors
Connect in under a minute. MCP (OAuth), full CLI, or a one-line agent skill.